Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/542806
Title: | Content Based Visual Information Retrieval Using Machine Learning Techniques |
Researcher: | Yashaswini, D K |
Guide(s): | Karibasappa, K |
Keywords: | Engineering Engineering and Technology Engineering Electrical and Electronic |
University: | Visvesvaraya Technological University, Belagavi |
Completed Date: | 2022 |
Abstract: | Content Based Image Retrieval (CBIR) model is helpful to retrieve images from database which are similar to the query image. The CBIR system has wide range of applications such as mining, artificial intelligent, and computer vision. The CBIR system is also used in the medical domain for diagnosis and decision making in treatment for patients. Effective retrieval method is required for large database images in the system. In the first work, dual phase model is used for improving the retrieval performances for CBIR system. The developed models performance in CBIR system were evaluated on COREL and Wang datasets. Normalization is applied in the input images to improve the visibility level of images. The Alex-Net Convolutional Neural Network (CNN) and colour moment methods were used for extracting the features from an input image. The semantic space between the extracted values is reduced using the combination of low- and high-level features that significantly improves the retrieving performance. Among the query image features and database images, the distances are measured based on the Manhattan distance. The dual phase model has higher performance in CBIR system in terms of f-measure, recall and precision. The proposed method shows the improvement of precision value of 0.43 and recall value of 0.06 in CBIR system compared to existing methods of Bi-layer system, Euclidean distance of spatial and frequency domain and colour histogram with local directional pattern. In second work, integration of Alex-Net CNN model, Local Optima Oriented Pattern (LOOP), and Grey Level Co-occurrence Matrix (GLCM) were proposed for CBIR system. The multi-space randomization and collaboration of the integration method for retrieval of semantic system. The segmentation of the foreground and the background image objects are performed using the super pixel based segmentation. The features from the objects are extracted by using the segmented region for balancing the subspace. It performs the process of randomization which partitions the mul |
Pagination: | 77 |
URI: | http://hdl.handle.net/10603/542806 |
Appears in Departments: | Department of Electrical and Electronics Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
01_title.pdf | Attached File | 62.88 kB | Adobe PDF | View/Open |
02_prelim pages.pdf | 134.55 kB | Adobe PDF | View/Open | |
03_content.pdf | 34.82 kB | Adobe PDF | View/Open | |
04_abstract.pdf | 4.78 kB | Adobe PDF | View/Open | |
05_chapter 1.pdf | 312.11 kB | Adobe PDF | View/Open | |
06_chapter 2.pdf | 93.14 kB | Adobe PDF | View/Open | |
07_chapter 3.pdf | 627.46 kB | Adobe PDF | View/Open | |
08_chapter 4.pdf | 647.84 kB | Adobe PDF | View/Open | |
09_chapter 5.pdf | 464 kB | Adobe PDF | View/Open | |
10_annexures.pdf | 81.52 kB | Adobe PDF | View/Open | |
80_recommendation.pdf | 11.77 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: